Neutron star collisions as a heavy element source

[phyzguy]

There was a lot of discussion after the recent observation of the merger of two neutron stars about whether or not these events are the source of the heavier elements. See this thread, for example. This recent paper has some new analysis. Especially interesting is Figure 10, that I've reproduced below. The paper's estimate of the abundance of the heavy elements is good match to the measured abundance of those elements here on Earth. It's looking more and more likely that NS-NS mergers are the source of those elements.

 

[anorlunda]

From the abstract of that paper

The rapid neutron-capture process, or r-process, is known to be fundamental for explaining the origin of approximately half of the A > 60 stable nuclei observed in nature.

Half is a lot. I'm curious. What mechanisms distribute those heavy elements throughout the galaxy?

 

[snorkack]

What was the speed at which the heavy elements of neutron star collision were propelled away?

 

[JMz]

From the abstract of that paperHalf is a lot. I'm curious. What mechanisms distribute those heavy elements throughout the galaxy?

Presumably kilonovae, as with the 2017 LIGO/VIRGO event.

 

[stefan r]

What was the speed at which the heavy elements of neutron star collision were propelled away?

Higher than the escape velocity. I would have expected a spray with a range of velocities. This paper says 0.25c for the blue component and 0.15 ±0.3c for the red component.

 

[alantheastronomer]

Neutron star collisions are way too rare an event to have produced all the observable heavy elements in the galaxy. Most, if not all, the heavy elements are undoubtedly produced in supernovae explosions, whose abundances are also a good match. See the book "Supernovae and Nucleosynthesis" by Arnett.

 

[stefan r]

Neutron star collisions are way too rare an event to have produced all the observable heavy elements in the galaxy. Most, if not all, the heavy elements are undoubtedly produced in supernovae explosions, whose abundances are also a good match. See the book "Supernovae and Nucleosynthesis" by Arnett.

That was published in 1996. Could not have included recent observations of binary neutron star mergers. Ligo did not exist.

 

[phyzguy]

Neutron star collisions are way too rare an event to have produced all the observable heavy elements in the galaxy. Most, if not all, the heavy elements are undoubtedly produced in supernovae explosions, whose abundances are also a good match. See the book "Supernovae and Nucleosynthesis" by Arnett.

Can you back up these statements with calculations of the rate of NS-NS mergers and how much heavy elements they produce? I doubt it, since people who have done these calculations disagree with you.

 

[JMz]

Neutron star collisions are way too rare an event to have produced all the observable heavy elements in the galaxy. Most, if not all, the heavy elements are undoubtedly produced in supernovae explosions, whose abundances are also a good match. See the book "Supernovae and Nucleosynthesis" by Arnett.

My understanding is that models show that SNs alone don't explain enough of the neutron-rich elements, like gold (even though they do well with the intermediate-mass elements, like iron).

 

[rootone]

My understanding is that models show that SNs alone don't explain enough of the neutron-rich elements, like gold (even though they do well with the intermediate-mass elements, like iron).

Iron can be formed in the most massive normal stars. although at that point the star doesn't have much of a future.
Not all Iron/Nickel producing stars necessarily go to supernova though.

 

[stefan r]

Iron can be formed in the most massive normal stars. although at that point the star doesn't have much of a future.
Not all Iron/Nickel producing stars necessarily go to supernova though.

Type 1a supernovas create a different mix of elements from type II and type 1b.

Type 1a SN produce a lot of nickel and iron. Massive stars create iron by burning lighter elements. Massive star cores can not explain abundances seen in the Milky Way because the center of a star's core stays inside the neutron star.

 

[alantheastronomer]

There's still a lot of uncertainty involved in yields due to supernovae, but models involving homologous cores, where, after neutrons are produced by protons absorbing electrons, there is no longer any gas pressure and the inner core collapses essentially at free fall, the outer core falls at the speed of sound and bounces off the inner core producing a shock wave, could possibly explain the heavier elemental abundances. This homologous core collapse scenario might account for the observations of neutron stars with masses of 0.8-0.9 solar masses.

 

[JMz]

How much nucleosynthesis does the outer core experience during the bounce, in that scenario?

As I understand it, the bounce is the only time that can happen, and the outer core is the only part with weight A >~ 50 that escapes the star (and maybe not most of that). So adding neutrons to to the outer core seems like the only way the SN can put neutron-rich nuclei into the ISM. Even then, I'd imagine most of the detritus (after radioactive decay) is only modestly enhanced in neutrons, so it wouldn't add much for, say, A > 100. True? Or do the simulations show a more favorable outcome?

 

[Chronos]

One thing that comes to mind: Since NS kicks, among other things, suggest supernova explosions can be highly asymmetric, it is not difficult to envision this leading to unusually rich pockets of heavy element forming in the ejecta. It would also facilitate their dispersal. I share the view that NS mergers are too rare to sufficiently enrich the cosmos to any meaningful extent. On the other hand, there is never a shortage of newborn NS stars to fall back upon. Perhaps there are some heavy elements that demand extraordinary energies to form in meaningful abundance, but, a more generic explanation appears probable for local elemental abundance. .

 

[alantheastronomer]

The paper that Phyzguy at the beginning of this thread refers to - The Decompression of the Outer Neutron Star Crust and R-Process Nucleosynthesis by Goriely,Chamel,Janka, and Pearson - is not unique to Neutron Star-Neutron Star collisions. Indeed it is identical to a homologeous core collapse with bounce shock albeit with a different geometry, and so the elemental yields should be expected to be identical for the two scenarios.

 

[phyzguy]

I share the view that NS mergers are too rare to sufficiently enrich the cosmos to any meaningful extent. .

You and @alantheastronomer have both made statements of this type, but I haven't seen any references or calculations to back up these statements, which makes them personal speculations. In addition to the paper I referenced in the OP, you can also look at this paper from the LIGO collaboration. Both make detailed estimates of the rate of NS-NS mergers, and come to the conclusion that NS-NS mergers happen at a rate sufficient to explain the R process elements. I've included a figure from this latter paper below. Of course, the rate is still uncertain, but I suspect that after another 5-10 years of LIGO/VIRGO observations it will be clear that there are more than enough of these events.

 

[phyzguy]

The paper that Phyzguy at the beginning of this thread refers to - The Decompression of the Outer Neutron Star Crust and R-Process Nucleosynthesis by Goriely,Chamel,Janka, and Pearson - is not unique to Neutron Star-Neutron Star collisions. Indeed it is identical to a homologeous core collapse with bounce shock albeit with a different geometry, and so the elemental yields should be expected to be identical for the two scenarios.

How does a core bounce eject neutron star material? Isn't this material tightly bound to the NS which is forming? I can see it ejecting material, but how does eject material which has been compressed to NS densities?

 

[alantheastronomer]

Let me take your latest post first: When I say core bounce, it refers to an inner core of about 0.8 solar masses which has formed a neutron star by, first freefalling due to a loss of pressure from electron capture onto protons that formed the neutrons, then suddenly stiffening due to the neutrons having to obey the Pauli Exclusion Principle. Then, the outer core, also formed by electron capture, falls, supersonically onto the inner core, bouncing off it causing a shock wave that diffuses the neutrons into the overlying layers of the stellar material which produces the r-process elements. This isn't the whole story however; The best computer simulations have yet to successfully produce a supernova explosion, so there is some as yet unknown mechanism that is responsible for supernovae explosions. One of the authors of the paper you originally referenced, Janka, has produced models that find the outer shockwave is strengthened when convection of the overlying material is taken into account, but we still have a long way to go.
Now for your former post: No one is arguing that the models aren't a remarkably accurate fit to terrestrial (and galactic) r-process abundances. On the left hand of the graph you provided, you'll find the vertical axis is labeled in Gigaparsecs. The Milky Way is only 100 thousand light years in diameter, so what the authors are referring to is a large fraction of the observable universe. The neutron star merger observed by LIGO was thought to be 150 million light years away, and so far there have been no gravitational waves or short gamma ray bursts observed within the milky way galaxy, even in Globular Clusters where it's thought that neutron stars sink to the center and x-ray sources have been observed.
As a quick contrast, it's been estimated that there are about 4 million neutron stars presently in the galaxy. Only a fraction of them will be in a double binary. On the other hand, using what's called the Initial Mass Function (IMF) of main sequence (hydrogen core burning) stars ( the IMF is the number of stars of a specific mass, low mass stars are more numerous than high mass stars) and a total number of stars in the galaxy of 200 billion (an EXTREMELY conservative estimate) we find a total number of stars that are massive enough to end in a supernova explosion of 4 billion. It's also been estimated that the number of supernova explosions in the galaxy is about 1 or 2 every hundred years, while the number of neutron star collisions is speculated to be 1 or 2 a year for the entire universe.
Does this clear things up?

 

[phyzguy]

It's also been estimated that the number of supernova explosions in the galaxy is about 1 or 2 every hundred years, while the number of neutron star collisions is speculated to be 1 or 2 a year for the entire universe. Does this clear things up?

No, I'm afraid this doesn't clear things up. Where does your estimate of NS-NS mergers of 1-2 per year for the entire universe come from? There are approximately 10^11 galaxies in the observable universe, so your estimate of NS-NS mergers is about 10^-11 per galaxy per year. Many people have done these estimates, and I won't list them all here, but almost all of these estimates have been in the range of 1-100 per galaxy per Myr, which is 10^-4 - 10^-6 per galaxy per year, which is 10^5-10^7 times greater than your estimate. Let me ask you two questions:

(1) If the rate is 10^-11 per galaxy per year, what would be the odds that LIGO, which was sensitive to around 10^6 galaxies, would detect a NS-NS merger in a few months of observing? Did they just get really really really lucky?

(2) If the rate is 10^-11 per galaxy per year, how can it be that even in the small portion of the Milky Way that we have surveyed, we already have found six verified NS-NS binaries, four of which will merge in the next few hundred million years? (See below)

Perhaps your confusion lies in calling these events "collisions". These are not NS wandering through space that randomly collide. These are binary star systems which evolve into binary NS, then spiral in and merge due to the emission of gravitational waves.

 

[alantheastronomer]

I'm sorry, you're right. Instead of "entire universe" I should have said "within the observational limits of LIGO". The estimate that you quote is the correct one. By "collisions" I still mean binary neutron star mergers, not random interactions. But still, given the estimated merger rates that you site there are about 10,000 more supernovae explosions than neutron star mergers per million years. Given that the yields from the two different processes are practically identical, it means that most of the r-process nuclei are most likely produced by supernovae explosions. The scarcity of mergers as compared to supernovae makes this most likely the case.

 

[phyzguy]

OK, now I understand better where you are coming from. Certainly there are many more supernovae than NS mergers. I think it all hinges on your statement that "Given that the yields from the two different processes are practically identical..." I don't think anybody knows at present whether this is true or not. But several people (like the paper I linked in post #16) have done the analysis and concluded that the production rate of r-process elements from NS mergers is sufficient to explain all of the existing r-process elements. If this is true, and if supernovae produce the same yield of r-process elements and are 1000's of times more numerous, then there should be far more of the r-process elements than we see. So either the estimates done of r-process elements from NS mergers are way off, or supernovae must produce these elements at a much lower rate than NS mergers. Time will tell which is the case.

 

[Chronos]

Narrowing down the binary neutron star merger rate is a vital step in determining if these events are a viable candidates for heavy element enrichment of the MW.. It is relevant to take into account that binary NS merger is one of the few environments believed even capable of supporting r-process nucleosynthesis. So, it is perfectly understandable detecting such an event would be of immense interest The number of related papers published is no coincidence. The two big questions relate to: 1] them amount of heavy nuclei liberated; and, 2] mergers frequency. This paper appears to cover the first count, but, falls a little short on the second.

The lack of well founded alternatives to NS mergers as a source of r-process elements appears to offer circumstantial encouragement for fine tuning projections of r-process element output and frequency of NS mergers. This paper, https://arxiv.org/pdf/1710.02142.pdf, points out that r-process contributions from single massive stars may still be necessary by noting

"Observations of lowest metallicity stars in our Galaxy and (ultra-faint) dwarf galaxies show substantial ”pollution” by r-process elements, indicating a production site with a low event rate and consistent high amount of r-process ejecta in order to explain solar abundances. This is also underlined by the large scatter of Eu/Fe (Eu being an r-process element and Fe stemming from CCSNe at these low metallicities) seen in the earliest stars of the Galaxy, indicating that in a not yet well mixed interstellar medium the products of regular CCSNe and these rare events vary substantially."

 

[stefan r]

... a shock wave that diffuses the neutrons into the overlying layers of the stellar material...

Is "diffusion" the right term? In chemistry particles diffuse by moving in random walks. A bounce is directed and not very random. It is not clear to me how neutrons typically move.

 

[alantheastronomer]

Is "diffusion" the right term? In chemistry particles diffuse by moving in random walks. A bounce is directed and not very random. It is not clear to me how neutrons typically move.

Let me clarify - If you're uncomfortable with the term "diffuses", shall I rather say "mixes" or "convects" or "collides"?

 

[eachus]

There is another factor involved. As more kilonova are observed with LIGO it should be possible to determine two numbers. First the minimum total mass, at collision, required to form a kilonova. Then the mass fraction which is lost in the last few hundred years before the collision. Unfortunately, the latter may not be a single number. It may depend on the magnetic fields of each neutron star. (Pulsars have very high magnetic fields, magnetars much higher.) As the stars get close, the fields should align. (Well one north pointing up, the other pointing down.) Now both baryonic and neutron star matter flow from pole to pole. The magnetic field lines are separated more widely midway between the stars than at the poles. The matter here becomes a plasma that is controlled more by electromagnetism than gravity. So the mass loss could become substantial. (I'm qualified to be part of a team simulating this on a supercomputer. But first physicists are going to have to figure out what mix of nucleii is hoovered up from the neutron stars surface. Oh, and what current, if any, will be flowing through the plasma. And anyway, I'm retired now. ;-)

 

[JMz]

Nice idea. But what fraction of NS's are pulsars? More precisely, for what fraction of NS mergers would magnetism be an important consideration?

 

[Chronos]

The ratio of pulsars vs neutron stars is challenging thing to figure given pulsars are the easiest way to positively identify a neutron star.

 

[JMz]

Yeah. But X-rays are another way to find NSs, which might give a handle on the ratio: imprecise, but even factor-of-2 accuracy would be plenty at this point.

BTW:

Now both baryonic and neutron star matter flow from pole to pole.

Presumably neutrons would not participate in this flow, as they are electrically neutral. At the NS surfaces, they will decay into protons & electrons, which can.

 

[Chronos]

Aside from pulsars/magnetars, neutron stars have only been found via x ray emissions in a few cases involving supernova remnants. See https://imagine.gsfc.nasa.gov/science/objects/neutron_stars1.html for discussion.

 

[JMz]

A good point, for specific detections. I was thinking probabilistically: If we know the approximate white dwarf/NS/black hole ratio from theory, and we detect some number of compact X-ray sources in, say, the MW as a whole (or at least in some large region around us), then we can estimate the number of NSs there and compare to the number of pulsar detections to estimate the fraction of NSs that are pulsars.

 

[Chronos]

Yes, you can measure the stochastic x ray background flux vs point source contributions and probabilize the fraction attributable to unresolved point sources as discussed here https://eud.gsfc.nasa.gov/Volker.Beckmann/school/download/JKnodlseder_Diffuse.pdf. This looks like a good way to humor check stellar evolution modeling.

 

[Chronos]

This is why we keep coming back. I never seem quite able to manage leaving here dumber than I was upon arrival.

 

[JMz]

I think there's a lot of that going on. ;-)

 

[alantheastronomer]

Yeah. But X-rays are another way to find NSs, which might give a handle on the ratio: imprecise, but even factor-of-2 accuracy would be plenty at this point.

BTW:

Presumably neutrons would not participate in this flow, as they are electrically neutral. At the NS surfaces, they will decay into protons & electrons, which can.

Neutrons don't have an electric charge, it's true, but they do have a magnetic moment on the same order of magnitude as the proton and so would be affected significantly by the neutron stars' powerful magnetic fields. In fact, it was the neutron's magnetic moment that gave the first hint that the neutron had some sort of internal structure.
But let me clarify another point - there's no reason to think that as the neutron stars get closer that their magnetic fields would force them to flip upside down. The magnetic force is very short range, and even though very powerful still overshadowed by gravitational attraction, although interaction of their magnetic fields might alter the neutron stars dynamics appreciably.
One more thing, from this thread it appears that you seem to think that there are pulsars and then there are neutron stars with no appreciable magnetic fields. All neutron stars have magnetic fields and are pulsars. The only reason we can't observe them all is that not all of them are aligned with our line of sight.

 

[JMz]

Neutrons don't have an electric charge, it's true, but they do have a magnetic moment on the same order of magnitude as the proton and so would be affected significantly by the neutron stars' powerful magnetic fields. In fact, it was the neutron's magnetic moment that gave the first hint that the neutron had some sort of internal structure.

True. I took the previous post to be focused on the currents, only.

...One more thing, from this thread it appears that you seem to think that there are pulsars and then there are neutron stars with no appreciable magnetic fields. All neutron stars have magnetic fields and are pulsars. The only reason we can't observe them all is that not all of them are aligned with our line of sight.

[If you are still addressing my post at this point, rather than @Chronos's, to which I was replying:]
Agreed, though my comment was only about X-rays as an alternative mode of discovery: The pulsar behavior might be unobservable because of alignment or otherwise, whereas looking for X-rays would also be unsuccessful for some (perhaps most) NSs but would be subject to a different set of selection effects than pulsar behavior.